Accepted Manuscript || Eur J Public Health. 2020. Oxford University Press

Home , University of Oxford

Accepted Manuscript || Eur J Public Health. 2020. Oxford University Press. doi: http://dx.doi.org/10.1093/eurpub/ckaa150 University Press. doi: 2020. Oxford Health. Eur J Public Accepted || Manuscript COVID-19 mortality: a complex interplay of sex, gender and ethnicity

Nazrul Islam, MBBS, PhD1,2 Kamlesh Khunti, MD, PhD, FMedSci3 Hajira Dambha-Miller, MRCGP, PhD2,4 Ichiro Kawachi, MB.ChB, PhD5 Michael Marmot, MBBS, PhD, FRCP6

1 Clinical Trial Service Unit and Epidemiological Studies Unit (CTSU), Nuffield Department of Population Health, University of Oxford, Oxford, UK 2 Medical Research Council Epidemiology Unit, University of Cambridge, Cambridge, UK 3 Leicester Diabetes Centre, University of Leicester, Leicester, UK 4 Department of Primary Care and Population Health, University of Southampton, Southampton, UK 5 Harvard T.H. Chan School of Public Health, Harvard University, Boston, MA, USA 6 Institute of Health Equity, Department of Epidemiology and Public Health, University College London, London, UK

Corresponding author: Nazrul Islam, MBBS, MSc, MPH, PhD Clinical Trial Service Unit and Epidemiological Studies Unit (CTSU) Nuffield Department of Population Health, Big Data Institute University of Oxford, Oxford, UK Email: [email protected]

Citation:

Islam N, Khunti K, Dambha-Miller H, Kawachi I, Marmot M. COVID-19 mortality: a complex interplay of sex, gender and ethnicity. Eur J Public Health. 2020. doi:10.1093/eurpub/ckaa150 Accepted Manuscript || Eur J Public Health. 2020. Oxford University Press. doi: http://dx.doi.org/10.1093/eurpub/ckaa150 University Press. doi: 2020. Oxford Health. Eur J Public Accepted || Manuscript Several studies have reported a higher rate of COVID-19 mortality in men.[1–3] A higher rate of COVID-19 mortality has also been reported in Black, Asian and minority ethnic (BAME) groups,[3–5] especially among healthcare providers.[6]

The exact reasons for these disparities are not known but may be due to differential susceptibility based on biological sex,[7] as well as gender differences in health behaviors (e.g. smoking) giving rise to differences in comorbidities (e.g., cardiovascular disease) that increase the risk of COVID-19 mortality in men.[8] However, there are social influences that could influence gender differences in exposure and infection; e.g., women are more likely to be involved in service sector work/healthcare; men are more involved in other high risk jobs such as drivers.[3,8] In regards to ethnic differences, people from BAME background may be more likely to be in frontline, exposed, jobs; they may be more likely to live in crowded multi- generation households making it challenging to maintain physical distancing from elderly family members.[9]

In the context of gender and ethnic differences in COVID-19 mortality, additional important policy-related issues include our understanding of (i) whether there are ethnic variations in COVID-19 mortality in men and women, (ii) whether there is heterogeneity in gender differences within individual ethnic groups, and (iii) whether we could identify some factors that may help explain these disparities, if any.

Most studies of COVID-19 mortality have statistically “adjusted” for factors (e.g., socioeconomic deprivation) that may potentially help to explain gender and ethnic disparities.[10] While often necessary, these adjustments are seldom sufficient in explaining the full spectrum of the disparities. Frequently, we do not have complete information on the causal pathways. Many known or hypothesised factors that could account for gender/ethnic disparities in health are not readily available in routine health records. Methodological and analytic approaches may also affect these conclusions.

For example, in the context of gender, ethnicity, and COVID-19 mortality, if gender is “adjusted” in a statistical model, it assumes that the ethnicity-specific risk estimates are fixed in men and women i.e., it renders invisible any heterogeneity in gender differences within individual ethnic groups. Similarly, if ethnicity is adjusted, the model assumes that the gender-specific risk estimates are fixed in different ethnic groups i.e., it masks any ethnic Accepted Manuscript || Eur J Public Health. 2020. Oxford University Press. doi: http://dx.doi.org/10.1093/eurpub/ckaa150 University Press. doi: 2020. Oxford Health. Eur J Public Accepted || Manuscript variation in men and women. Therefore, just because an association is reported as ‘adjusted’, it is no panacea.

We elaborate these issues further drawing on results from three recent reports published using the UK data. The OpenSAFELY study reported that the COVID-19 mortality risk in men was twice as high compared to women.[1] Since this study adjusted for both gender and ethnicity, it assumes that the increased risk of COVID-19 mortality in men is fixed regardless of ethnicity, and that the ethnicity-specific risks are also fixed in men and women. However, the results from the UK Office for National Statistics (ONS) showed that there was substantial heterogeneity in the risks of COVID-19 mortality in different ethnic groups, both in men and women.[4]

However, the ONS results do not allow the comparison of risks in men and women within individual ethnic groups because the study estimated the risk in men against White men, and that in women against White women. Therefore, to be able to directly compare the gender- specific risks in COVID-19 mortality in individual ethnic groups, we need to precisely know the gender difference in COVID-19 mortality in the reference (White) population. The unadjusted risk was approximately 1.5 times higher in White men (compared to White women) in the ONS study,[4] while the age-adjusted risk ratio was approximately 2.0 in the recent study by Public Health England (PHE).[3] Applying these estimates (1.5 and 2.0) to the ONS regression results, we find that the increased risk in men varies considerably across the ethnic groups (depending on the ONS statistical models, the risk in men varies between 1.3 and 3.5 times that in women in different ethnic groups) (Figure 1). These findings are suggestive of an ‘effect modification’, which also mandates presentation of these results stratified by the effect modifier instead of adjusting for them in the regression models.

After adjusting for a range of socioeconomic and structural factors, the ONS study showed that a considerable portion of ethnic variability could be explained by socioeconomic and structural factors (e.g., deprivation, household composition, regional variabilities). However, we do not know if this is true for gender differences within individual ethnic groups. In the context of gender differences in COVID-19 mortality, it will be invaluable to understand whether the differences in men and women could potentially be explained by determinants related to biological sex or to social factors (gender). These findings will help shape public health policies on the prevention and treatment of COVID-19. Accepted Manuscript || Eur J Public Health. 2020. Oxford University Press. doi: http://dx.doi.org/10.1093/eurpub/ckaa150 University Press. doi: 2020. Oxford Health. Eur J Public Accepted || Manuscript A growing body of research is attempting to examine the relationship between sex hormones and COVID-19 susceptibility, which could potentially help explain the sex (biological) differences.[7] Future studies should explore the effects of additional factors, including (but not limited to) pattern, sequence, and duration of multimorbidity, on COVID-19 susceptibility/mortality within the context of individual ethnic groups to disentangle these issues.

Funding: No specific funding for this project. NI receives salary support from the Nuffield Department of Population Health (NDPH), University of Oxford. KK acknowledges support from the National Institute for Health Applied Research Collaboration East Midlands (NIHR ARC EM) and the NIHR Leicester Biomedical Research Centre. The views expressed in this article are those of the authors and not necessarily those of the entities the authors are affiliated with and/or supported by.

Conflicts of interest: KK is Director of the Black and Minority Ethnic Centre, NIHR Applied Research Collaborations (ARC) East Midlands. Other authors declare no conflicts of interests to declare.

References:

1 Williamson E, Walker AJ, Bhaskaran KJ, et al. OpenSAFELY: factors associated with COVID- 19-related hospital death in the linked electronic health records of 17 million adult NHS patients. medRxiv Published Online First: 2020. doi:10.1101/2020.05.06.20092999

2 Docherty AB, Harrison EM, Green CA, et al. Features of 20 133 UK patients in hospital with covid-19 using the ISARIC WHO Clinical Characterisation Protocol: prospective observational cohort study. BMJ 2020;369:m1985. doi:10.1136/bmj.m1985

3 Public Health England. Disparities in the risk and outcomes of COVID-19. London, UK: : Public Health England 2020.

4 Office for National Statistics. Coronavirus (COVID-19) related deaths by ethnic group, England and Wales. London, UK: : Office for National Statistics 2020.

5 Platt L, Ross W. Are some ethnic groups more vulnerable to COVID-19 than others? London, UK: : The Institute for Fiscal Studies 2020.

6 Majeed A, Molokhia M, Pankhania B, et al. Protecting the health of doctors during the COVID-19 pandemic. Br J Gen Pract 2020;70:268–9. doi:10.3399/bjgp20X709925

7 Wadman M. Why coronavirus hits men harder: sex hormones offer clues. Science Published Online First: 3 June 2020. doi:10.1126/science.abd1288 Accepted Manuscript || Eur J Public Health. 2020. Oxford University Press. doi: http://dx.doi.org/10.1093/eurpub/ckaa150 University Press. doi: 2020. Oxford Health. Eur J Public Accepted || Manuscript 8 Burström B, Tao W. Social determinants of health and inequalities in COVID-19. Eur J Public Health 2020;:ckaa095. doi:10.1093/eurpub/ckaa095

9 Office for National Statistics. Families and households: UK Ethnicity facts and figures. London, UK: : Office for National Statistics 2019.

10 Hernán MA, Clayton D, Keiding N. The Simpson’s paradox unraveled. Int J Epidemiol 2011;40:780–5. doi:10.1093/ije/dyr041

Figure 1: Adjusted odds ratio estimates of COVID-19 mortality in men compared to women in the United Kingdom

Models: 1- adjusted for Age; 2- Model 1 + region, rural/urban; 3- Model 2 + IMD decile; 4- Model 3 + household composition; 5- Model 4 + socio-economic status; 6- Model 5 + health status. More details are available at: https://www.ons.gov.uk/peoplepopulationandcommunity/birthsdeathsandmarriages/deaths /articles/coronavirusrelateddeathsbyethnicgroupenglandandwales/2march2020to10april202 0